GEO vs. Agent Selection: Citations Aren't the Same as Getting Picked
A wave of tools will tell you how often AI mentions your brand. Profound, Ahrefs Brand Radar, a dozen others — they track citations and share of voice in AI answers. That category has a name: GEO, generative engine optimization. It's useful. It also measures the wrong finish line.
Citations measure attention. Selection measures revenue.
A citation means an AI mentioned you. A selection means an AI chose you — wrote your import, called your API, shipped your SDK into the user's codebase. The first is awareness. The second is the transaction. You can lead every "best tools for X" answer and still lose every actual build, because when the agent stops talking and starts coding, it picks one tool, and prose ranking has surprisingly little to do with which one.
Where they diverge in practice
Watch what happens to incumbents. In search, the household-name incumbent gets cited constantly — it's in every comparison. But on the unbranded task "add search to my app," a leaner challenger edges ahead: the challenger wins roughly 17% of trials to the incumbent's 14%. Citations said the incumbent was winning. Selection said otherwise. Only one of those shows up in your signups.
The same pattern hits transactional email, where the most-marketed name isn't the most-picked one. If you'd optimized purely for citations, you'd have declared victory while quietly losing the builds.
What to actually optimize
GEO and Agent SEO share a foundation — corpus presence helps both — so this isn't either/or. But the scoreboard matters. Optimizing for citations rewards being talked about. Optimizing for selection rewards being the right answer when an agent commits. Those produce different roadmaps: more thought-leadership content versus a tighter quickstart, cleaner machine-readable docs, and winning the unbranded prompt.
Measure the choice
If you only get to instrument one number, instrument the one that ends in your package being imported. That's Pick Rate, and you can see how it diverges from the citation story on the live leaderboards. Citations tell you you're in the conversation. Pick Rate tells you whether you won it.
FAQ
What's the difference between GEO and Agent SEO?
GEO (generative engine optimization) optimizes for being cited in AI-generated answers. Agent SEO optimizes for being selected when an agent acts. One measures presence in prose; the other measures the choice that ends in your package being imported.
Do I need both?
They overlap on corpus presence, so good GEO helps awareness. But a strong citation profile with a weak Pick Rate is common and costly — you're talked about and not chosen. If you can only measure one thing, measure the selection.
See your Pick Rate
Check how often AI agents pick your tool — free, no account needed.
Check your tool →Keep reading
Serve Both: /index.md and Accept: text/markdown
Content negotiation is the correct way to serve markdown to agents. It is also the one most agents don't use. Cloudflare tested seven; three sent the header. Here's why you need the ugly URL fallback too, and why that stays true no matter how good the new discovery standards get.
We Scored 124 Dev Tools on Agent Readiness. It Barely Predicts Who Agents Pick.
We ran Cloudflare's Agent Readiness score against every tool on the Pickrate leaderboards, then joined it to measured Pick Rate. The correlation is 0.19. Stripe scores 4 out of 16 and wins 80% of payments trials. Telnyx scores 14 out of 16 and wins nothing.